cluster lock

[Nabil Boussetta]

we have an MC/service guard cluster on two node (two rp8400).
the cluster lock is on eva 5000.

we have deleted the vg of the cluster lock on two nodes.
after that. we have created another virtual disk on the eva. we have presented this disk to the nodes with the old device name c16t0d0. and we have recreated the vg .

we still have an error on the syslog all the hours:

Feb 6 08:54:53 CENTRAL2 cmcld: WARNING: Cluster lock on disk /dev/dsk/c16t0d0 is missing!
Feb 6 08:54:53 CENTRAL2 cmcld: Until it is fixed, a single failure could
Feb 6 08:54:53 CENTRAL2 cmcld: cause all nodes in the cluster to crash
Feb 6 08:54:54 CENTRAL2 cmclconfd[7854]: Updated file /var/adm/cmcluster/frdump.cmcld.3 for node CENTRAL2 (length = 512096).

Feb 6 09:56:08 CENTRAL2 cmcld: WARNING: Cluster lock on disk /dev/dsk/c16t0d0 is missing!
Feb 6 09:56:08 CENTRAL2 cmcld: Until it is fixed, a single failure could
Feb 6 09:56:08 CENTRAL2 cmcld: cause all nodes in the cluster to crash
Feb 6 09:56:09 CENTRAL2 cmclconfd[7854]: Updated file /var/adm/cmcluster/frdump.cmcld.4 for node CENTRAL2 (length = 512096).


what is missing ?

[melvyn burnard]

As you hav erecretaed the VG, you have destroyed the cluster informatin especially the fact that this is a cluster lock fdisc.
You need to either recover this, using vgcfgrestore IF you did vgcfgbackup AFTER you had made this the cluster lock disc, or you will need to obtain the cminitlock tool from your local HP Response Centre, or you will need to re-apply the cluster with thois vg activated on a node and the cluster DOWN

[paolo barila]

Hi,
it seems to me you didn't do the "cmapplyconf" command

# cmapplyconf -C /etc/cmcluster/

where is your file where you have:

FIRST_CLUSTER_LOCK_VG /dev/vglockdisk
FIRST_CLUSTER_LOCK_PV /dev/dsk/c....
VOLUME_GROUP /dev/vglockdisk

Remember also (before) to activate your lock disk:

# vgchange -a y /dev/vglockdisk

I understand didn't change but you still need to apply it to the new disk

Pablo

[Devender Khatana]

Hi,

The device for cluster lock was removed and was represented, but I do not think it was done using vgcfgrestore command. Was it?

The solution would be to reconfigure cluster lock structure on disk either by reconfiguring it having done vgcfgrestore through old map file or will have to reapply config.

Refer this thread for a similar problem
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=75245

HTH,
Devender

[Nabil Boussetta]

i have executed cmapplyconf after all changes
but the output of the command tells me that there is no change in the config.

i' havn't used vgcfgrestore to recreate the vg .

[paolo barila]

I did this sequence and it worked:

# cmhaltcl -fv # stop cluster!

# cmcheckconf -C /etc/cmcluster/cmclconf.ascii

# vgchange -a y /dev/vglockdisk

# cmapplyconf -C /etc/cmcluster/cmclconf.ascii

# cmruncl -v

[Nabil Boussetta]

ok

[Devender Khatana]

Hi,

It is important here that the vg should be activated while applying config. Is error repeating after reapplying config?

HTH,
Devender

[Stephen Doud]

I have seen cmapplyconf do nothing when the content of the cluster configuration file does not change.
If you have not changed the name of the cluster lock VG (FIRST_CLUSTER_LOCK_VG) or the paths to the cluster lock disk (FIRST_CLUSTER_LOCK_PV), then cmapplyconf will see no reason to affect any changes, and will do nothing.

In your case, it may be necessary to delete the cluster and re-apply it.
# cmhaltcl -f
# cmdeleteconf -f
# vgchange -a y
# cmapplyconf -C -P ....
# vgchange -a n